We are currently planning the activities within the Pufendorf Advanced Study Group Quantifying and Communicating Uncertainty: Challenges and Opportunities. This will be a year with plenty of opportunities to boost your Bayesian knowledge and experience. There will be courses and invited lectures. Info are continuously posted at the website of the ASG.

The goal of the Advanced Study Group at Pufendorf Institute in Lund is to advance the understanding of Bayesian methods of data analysis for quantification and communication of uncertainty. Uncertainty is inherent in the scientific process and especially crucial to quantify and communicate effectively when it affects health, safety, and the environment. To this end we propose an interdisciplinary effort aimed at determining how Bayesian methods can be used more effectively when evaluating competing propositions in the face of uncertainty. Special focus will not only be put on how Bayesian methods can help solve the restrictions of classical analysis methods, but also on how challenges of doing Bayesian data analysis, both practical and conceptual, can be addressed.

Members of this group are:

Jonas Björk (Department of Laboratory Medicine)
Rasmus Bååth (Department of Philosophy)
Johan Lindström (Centre for Mathematical Sciences)
Geoffrey Patching (Department of Psychology)
Kurt Petersen (Division of Risk Management)
Umberto Picchini (Centre for Mathematical Sciences)
Krzysztof Podgorski (Department of Statistics)
Nils-Eric Sahlin (Faculty of Medicine)
Ullrika Sahlin (Centre of Environmental and Climate Research)
Martin Stjernman (Department of Biology)

The summer school in Bayesian Analysis and Decision Theory at Lund University 2015 has now completed its physical week. Students are now working on their individual projects and we expect a thrilling web conference at late September.

All lectures and exercises were held in the Maths Building at Lund University. We got the opportunity to write on two walls of black boards using old fashioned chalks.

Monday 17^th

10.15 – 10.45 Course introduction by Ullrika Sahlin

10.45 – 14.00 Introduction to Bayesian Analysis by Johan Lindström

14.15 – 17.00 Bayesian Belief Networks I by Ullrika Sahlin

At the evening we had a welcome Pizza night at the Centre of Environmental and Climate Research at the Ecology Building including a quite geeky Bayesian Pub Quiz.

Tuesday 18^th

8.30 – 10.30 Introduction to Decision Theory by Matthias Troffaes

10.45 – 15.00 Bayesian Hierarchical Modelling I by Johan Lindström and Yann Clough

15.15 – 17.00 Probability, values and cautionary principles by Niklas Vareman

Wednesday 19^th

8.30 – 10.30 Bayesian Belief Networks II by Ullrika Sahlin

10.45 – 15.00 Bayesian Hierarchical Modelling II by Johan Lindström, Yann Clough and Ullrika Sahlin

15.15 – 17.00 Uncertainty in quantitative risk assessment I by Roger Flage

At the evening we went to the Malmö City Festival.

Thursday 20^th

8.30 – 11.00 Robust Bayesian Analysis by Matthias Troffaes and Ullrika Sahlin

11.15 – 14.00 Introduction and preparation of project ideas and the literature seminar

14.15 – 15.00 Uncertainty in quantitative risk assessment II by Roger Flage

15.15 – 17.00 Literature seminar by Niklas Vareman

At the evening some of us went to the city of Lund for some food and beer.

Friday 21^st

9.00  Presentation and feedback on projects ideas

Lunch at Ullrika’s favourite place at Finn Ut.

It was a great idea to combine BA with DT!

Now time for projects….

Our fabolous communication staff helped us to document the presentations and put them out on the web. Check this out!

Here is a summary of the workshop

Engaging stakeholders in research projects

Anna-Maria Jönsson (The interface between science and stakeholder interactions)talked about the outcome of a study they had made on how stakeholders and researchers perceived the stakeholder interactions and dialogues.

Susanna Bruzell (Integrating stakeholders in adaptation research – a case study of the Swedish forestry sector’s decision-making in a changing climate.) provided examples of the stakeholder dialogues in a larger research environment.

Johanna Alkan-Olsson (What does science say about how and why to interact with stakeholders in research projects) argued for the need to engage stakeholders in research.

Formulating and solving a multi-criteria decision problem

Ullrika Sahlin (The science of Multi-Criteria Decision analysis) gave an introduction to Decision theory, decision making under uncertainty and the formulation of complex decision problems.

Deniz Koca (The Role of Stakeholder Participation and Group Modelling in Developing Decision Support) showed how he in interaction with experts and stakeholders specify the system model upon which a decision analysis will be based.

Nils Cronberg (How we teach decision making in a netbased course in environmental science) presented how they work with teaching students methods to perform evidence based conservation.

Carin Nilsson (Adapting to an uncertain climate – lessons from practice – the CIRCLE-2 findings from Europe) picked some results from a larger study on methods to address and communicate uncertainty in climate adaptation.

Obtaining data from stakeholders

Stina Alriksson (Methods to measure stakeholder’s preferences) showed how she is obtaining data on how stakeholder’s value multiple objectives which can be further used in decision analysis but also to identify groups with common values.

General discussion covered issues like

We concluded that there is a need to build a shared competence and experience of scientific principles to address complex decision problems. We should be better to interact on this type of issues. Knowing the various competences among us on these methods can help knowledge exchange. We need to increase our capacity to utilize existing research projects to work towards managers and policy makers. A competence and resources to integrate decision theory in research and more directly work towards decision makers could be something that makes CEC different from traditional departments at the university.

Integrating decision making into science challenges our view and models of science. A discussion on how to meet the demand for science to provide decision support should be combined with a discussion on scientific integrity.

We should build up a capacity to not only transfer scientific knowledge but to also specify and develop the scientific methodology on doing so in parallel with knowledge transfer. There is e.g. a need for an alternative to evidence-based methods that fits into the type of applications we are working with.

We suggest making some effort to create contact points between research projects working with multi-criteria decision making or that has a goal to communicate uncertainty in their results. A second workshop of a similar kind is desired. Future topics could cover how to explain the development and use of scenarios and communication of uncertainty as part of the scientific method and not something that comes in when the results are there.

i.e. to be continued. / Ullrika

Introduction to Bayesian Hierarchical Modelling

– a pre-conference hands-on session

Ullrika Sahlin and Paul Caplat, Centre of Environmental and Climate Reserach

Bayesian modelling allows addressing complex problems with flexible model structure arranged in a nested or hierarchical way (hence the name BHM). We will demonstrate Bayesian updating using MCMC sampling in two commonly used open source software. For the sake of simplicity and generality we will introduce BHM using simple examples, increasing complexity (i.e. hierarchical nature) of the models step by step. Bayesian modelling allows addressing complex problems with flexible model structure arranged in a nested or hierarchical way (hence the name BHM).

We hope that the attendants will leave the workshop able to further explore the fun and usefulness of BHM.

Participants were asked to bring their own laptop. It was also possible to attend by looking over the shoulder of someone else.

The modelling will be performed in R with pre-made code that we put up on the conference website prior to the workshop. We recommended installing the programs before the workshop. We are running the examples in two types of software. Which one you choose is up to you, both have their advantages and they look very similar in the R-coding.

The examples we will go through are the following:

Bayesian updating: •The chance to get heads up in a tossing experiment using a Binomial model. •Bayesian linear regression and the influence of priors having little or lots of data.

Hierarchical modelling: •Bayesian linear modelling with a random effect. •Bayesian calibration of a process based model of population dynamics over time. •Bayesian calibration of the population dynamics model considering the error in observations using a Binomial model.